A design methodology

Abstract

It may seem ironical that a design methodology based on statistically defined inputs is to be preferred when complex wear processes are involved, but is of doubtful advantage for the much simpler stress-rupture mechanisms. The reason for this must be fully recognised. A lower level of design reliability is acceptable for wear because it can be countered by maintenance (when properly carried out), while maintenance can have no effect at all on the stress-rupture processes. Thus for aircraft structures, for which a very high reliability might be thought a sine qua non on account of safety, a full life design probability of fatigue failure as high as 1 in 1000 is considered adequate — though this is not of course the overall probability of failure. Consequently, for design involving wear processes only the accessible part of the probability density curve is generally required to provide the statistical input data. The superiority of design based on statistically defined inputs then becomes overwhelming — as one would expect. For stress-rupture an inaccessible part of the probability density curve is required for an accurate quantitative solution. In such circumstances statistical methods are not invalid, but they are impotent. So too are the empirical ‘factor’ methods. Techniques that claim to access the inaccessible are just crystal balls. Added to which, it has been shown that ‘factors’ are fundamentally flawed, in the case of both stress-rupture and wear, even when applied to the accessible part of the curve.